Why Open Loop Misses

Tell two identical motors to run at the same command and the robot may still curve.

Manufacturing variation, friction, load, wheel diameter, surface contact, and battery conditions do not disappear because software sent matching numbers.

Open-loop control assumes the command is enough.

Feedback control measures what happened and uses the difference.

Define the error

For a target value:

error = setpoint - measurement

If a wheel target is 2.0 m/s and measured speed is 1.6 m/s:

error = 0.4 m/s

The controller can use that error to adjust output.

Closed-loop picture

setpoint
   ↓
compare ←──────── measurement
   ↓ error             ↑
controller              |
   ↓                    |
actuator → mechanism → sensor

The sensor closes the loop.

Disturbance

Imagine the robot drives onto carpet. The load increases and speed drops.

An open-loop motor command remains the same.

A closed-loop speed controller sees larger error and can increase output, within available limits.

The phrase within available limits matters.

Feedback cannot create torque the motor, battery, driver, or mechanism does not have.

Failure mode: bad feedback

If the encoder direction is reversed, increasing motor output can make the measured error appear larger instead of smaller. The controller may drive harder in the wrong direction.

Feedback amplifies the quality of the information you feed it.

Draw two versions

For one robot mechanism, draw:

1. open-loop architecture;
2. closed-loop architecture.

Mark the disturbance you expect the closed loop to handle.

Then name one condition the feedback system still cannot solve.

Disturbances expose open-loop assumptions

Imagine commanding the same motor output in two trials.

Trial A occurs on a smooth floor with a fresh battery. Trial B occurs on carpet with a lower battery voltage.

If the controller never measures actual motion, it has no direct evidence that the two trials are different.

command 0.40 → motor → motion

That is an open-loop path.

A feedback system adds measurement:

target
  ↓
compare ← measured speed
  ↓
controller
  ↓
motor
  ↓
wheel
  └──── sensor ──── back

Now the controller can react to some disturbances because it can observe their effect.

Feedback does not remove physics

Suppose the target wheel speed is 2.0 m/s, but the motor and gearing can only produce 1.4 m/s under the current load. The controller may keep increasing its command, but feedback cannot create missing actuator capacity.

This is a key failure mode: control authority is limited.

Before adding feedback, identify:

• the target variable;
• the sensor that measures it;
• the actuator that can change it;
• the range over which the actuator has enough authority;
• the update rate;
• the safe behavior when the measurement becomes invalid.

Closed loop is not a magic upgrade. It is a relationship among measurement, decision, actuation, and physical capability.